{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 100,
  "global_step": 63,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 2.308793434500694,
      "epoch": 0.963855421686747,
      "grad_norm": 1.8339245319366455,
      "learning_rate": 1.9000000000000002e-06,
      "logits/chosen": 0.41484330565815475,
      "logits/rejected": 0.5456838045995205,
      "logps/chosen": -49.526001167297366,
      "logps/rejected": -59.49296364784241,
      "loss": 0.6891121387481689,
      "mean_token_accuracy": 0.22993013681843877,
      "num_tokens": 37803.0,
      "rewards/accuracies": 0.759375,
      "rewards/chosen": 7.2634290063433585e-06,
      "rewards/margins": 0.008121937591931782,
      "rewards/rejected": -0.008114674211174133,
      "step": 20
    },
    {
      "entropy": 2.277806546114668,
      "epoch": 1.9156626506024095,
      "grad_norm": 2.0258429050445557,
      "learning_rate": 3.900000000000001e-06,
      "logits/chosen": 0.4389253231148894,
      "logits/rejected": 0.5264429102267841,
      "logps/chosen": -48.951683044433594,
      "logps/rejected": -61.126885426195365,
      "loss": 0.61369309425354,
      "mean_token_accuracy": 0.22803395109463342,
      "num_tokens": 74911.0,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.004455907024219161,
      "rewards/margins": 0.17089249783113034,
      "rewards/rejected": -0.1664365911031071,
      "step": 40
    },
    {
      "entropy": 2.1899878133701374,
      "epoch": 2.8674698795180724,
      "grad_norm": 1.5958611965179443,
      "learning_rate": 5.9e-06,
      "logits/chosen": 0.36904452201936805,
      "logits/rejected": 0.4032783324493887,
      "logps/chosen": -48.9308927028994,
      "logps/rejected": -66.71900814394408,
      "loss": 0.40452375411987307,
      "mean_token_accuracy": 0.21395687834371493,
      "num_tokens": 112046.0,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.009511558775188802,
      "rewards/margins": 0.7412423389621928,
      "rewards/rejected": -0.731730774988102,
      "step": 60
    },
    {
      "epoch": 3.0,
      "eval_entropy": 2.0709543228149414,
      "eval_logits/chosen": 0.3332947482681086,
      "eval_logits/rejected": 0.30760469286591857,
      "eval_logps/chosen": -49.526841735839845,
      "eval_logps/rejected": -74.71841888427734,
      "eval_loss": 0.23298563063144684,
      "eval_mean_token_accuracy": 0.1670339286327362,
      "eval_num_tokens": 117000.0,
      "eval_rewards/accuracies": 1.0,
      "eval_rewards/chosen": -0.007996635138988495,
      "eval_rewards/margins": 1.3904076099395752,
      "eval_rewards/rejected": -1.3984042167663575,
      "eval_runtime": 0.8652,
      "eval_samples_per_second": 20.803,
      "eval_steps_per_second": 5.779,
      "step": 63
    }
  ],
  "logging_steps": 20,
  "max_steps": 63,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2301025416609792.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}